ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.67 by root, Sun Nov 29 21:35:06 2020 UTC

…		…
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
		11	#include <inttypes.h>
11		12
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
24	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	31	#endif
27		32
		33	// known major and minor types
		34	enum cbor_type
		35	{
		36	MAJOR_SHIFT = 5,
		37	MINOR_MASK = 0x1f,
		38
		39	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		40	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		41	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		42	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		43	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		44	MAJOR_MAP = 5 << MAJOR_SHIFT,
		45	MAJOR_TAG = 6 << MAJOR_SHIFT,
		46	MAJOR_MISC = 7 << MAJOR_SHIFT,
		47
		48	// INT/STRING/ARRAY/MAP subtypes
		49	LENGTH_EXT1 = 24,
		50	LENGTH_EXT2 = 25,
		51	LENGTH_EXT4 = 26,
		52	LENGTH_EXT8 = 27,
		53
		54	// SIMPLE types (effectively MISC subtypes)
		55	SIMPLE_FALSE = 20,
		56	SIMPLE_TRUE = 21,
		57	SIMPLE_NULL = 22,
		58	SIMPLE_UNDEF = 23,
		59
		60	// MISC subtype (unused)
		61	MISC_EXT1 = 24,
		62	MISC_FLOAT16 = 25,
		63	MISC_FLOAT32 = 26,
		64	MISC_FLOAT64 = 27,
		65
		66	// BYTES/TEXT/ARRAY/MAP
		67	MINOR_INDEF = 31,
		68	};
		69
28	// known tags	70	// known tags
29	enum cbor_tag	71	enum cbor_tag
30	{	72	{
31	// inofficial extensions (pending iana registration)	73	// extensions
		74	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
32	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	75	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
33	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	76	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
34	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	77	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
35	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	78	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
36	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	79	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
37	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
38	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	80	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39		81
40	// rfc7049	82	// rfc7049
41	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	83	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	84	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43	CBOR_TAG_POS_BIGNUM = 2, // byte string	85	CBOR_TAG_POS_BIGNUM = 2, // byte string
44	CBOR_TAG_NEG_BIGNUM = 3, // byte string	86	CBOR_TAG_NEG_BIGNUM = 3, // byte string
45	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	87	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46	CBOR_TAG_BIGFLOAT = 5, // array	88	CBOR_TAG_BIGFLOAT = 5, // array
47		89
48	CBOR_TAG_CONV_B64U = 21, // base64url, any	90	CBOR_TAG_CONV_B64U = 21, // base64url, any
49	CBOR_TAG_CONV_B64 = 22, // base64, any	91	CBOR_TAG_CONV_B64 = 22, // base64, any
50	CBOR_TAG_CONV_HEX = 23, // base16, any	92	CBOR_TAG_CONV_HEX = 23, // base16, any
51	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	93	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52		94
53	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	95	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	96	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	97	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	98	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58		100
59	CBOR_TAG_MAGIC = 55799 // self-describe cbor	101	CBOR_TAG_MAGIC = 55799, // self-describe cbor
		102	};
		103
		104	// known forced types, also hardcoded in CBOR.pm
		105	enum
		106	{
		107	AS_CBOR = 0,
		108	AS_BYTES = 1,
		109	AS_TEXT = 2,
		110	AS_INT = 3,
		111	AS_FLOAT16 = 4,
		112	AS_FLOAT32 = 5,
		113	AS_FLOAT64 = 6,
		114	// possibly future enhancements: (generic) float, (generic) string
60	};	115	};
61		116
62	#define F_SHRINK 0x00000001UL	117	#define F_SHRINK 0x00000001UL
63	#define F_ALLOW_UNKNOWN 0x00000002UL	118	#define F_ALLOW_UNKNOWN 0x00000002UL
64	#define F_ALLOW_SHARING 0x00000004UL //TODO	119	#define F_ALLOW_SHARING 0x00000004UL
65	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	120	#define F_ALLOW_CYCLES 0x00000008UL
		121	#define F_FORBID_OBJECTS 0x00000010UL
		122	#define F_PACK_STRINGS 0x00000020UL
		123	#define F_TEXT_KEYS 0x00000040UL
		124	#define F_TEXT_STRINGS 0x00000080UL
		125	#define F_VALIDATE_UTF8 0x00000100UL
66		126
67	#define INIT_SIZE 32 // initial scalar size to be allocated	127	#define INIT_SIZE 32 // initial scalar size to be allocated
68		128
69	#define SB do {	129	#define SB do {
70	#define SE } while (0)	130	#define SE } while (0)
…		…
89	typedef struct {	149	typedef struct {
90	U32 flags;	150	U32 flags;
91	U32 max_depth;	151	U32 max_depth;
92	STRLEN max_size;	152	STRLEN max_size;
93	SV *filter;	153	SV *filter;
		154
		155	// for the incremental parser
		156	STRLEN incr_pos; // the current offset into the text
		157	STRLEN incr_need; // minimum bytes needed to decode
		158	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94	} CBOR;	159	} CBOR;
95		160
96	ecb_inline void	161	ecb_inline void
97	cbor_init (CBOR *cbor)	162	cbor_init (CBOR *cbor)
98	{	163	{
…		…
102		167
103	ecb_inline void	168	ecb_inline void
104	cbor_free (CBOR *cbor)	169	cbor_free (CBOR *cbor)
105	{	170	{
106	SvREFCNT_dec (cbor->filter);	171	SvREFCNT_dec (cbor->filter);
		172	SvREFCNT_dec (cbor->incr_count);
107	}	173	}
108		174
109	/////////////////////////////////////////////////////////////////////////////	175	/////////////////////////////////////////////////////////////////////////////
110	// utility functions	176	// utility functions
111		177
…		…
137		203
138	// minimum length of a string to be registered for stringref	204	// minimum length of a string to be registered for stringref
139	ecb_inline int	205	ecb_inline int
140	minimum_string_length (UV idx)	206	minimum_string_length (UV idx)
141	{	207	{
142	return idx > 23	208	return idx <= 23 ? 3
143	? idx > 0xffU	209	: idx <= 0xffU ? 4
144	? idx > 0xffffU	210	: idx <= 0xffffU ? 5
145	? idx > 0xffffffffU	211	: idx <= 0xffffffffU ? 7
146	? 7	212	: 11;
147	: 6
148	: 5
149	: 4
150	: 3;
151	}	213	}
152		214
153	/////////////////////////////////////////////////////////////////////////////	215	/////////////////////////////////////////////////////////////////////////////
154	// encoder	216	// encoder
155		217
…		…
168	} enc_t;	230	} enc_t;
169		231
170	ecb_inline void	232	ecb_inline void
171	need (enc_t *enc, STRLEN len)	233	need (enc_t *enc, STRLEN len)
172	{	234	{
173	if (ecb_expect_false (enc->cur + len >= enc->end))	235	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
174	{	236	{
175	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	237	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	238	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177	enc->cur = SvPVX (enc->sv) + cur;	239	enc->cur = SvPVX (enc->sv) + cur;
178	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	240	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
184	{	246	{
185	need (enc, 1);	247	need (enc, 1);
186	*enc->cur++ = ch;	248	*enc->cur++ = ch;
187	}	249	}
188		250
		251	// used for tags, intregers, element counts and so on
189	static void	252	static void
190	encode_uint (enc_t *enc, int major, UV len)	253	encode_uint (enc_t *enc, int major, UV len)
191	{	254	{
192	need (enc, 9);	255	need (enc, 9);
193		256
194	if (len < 24)	257	if (ecb_expect_true (len < LENGTH_EXT1))
195	*enc->cur++ = major | len;	258	*enc->cur++ = major | len;
196	else if (len <= 0xff)	259	else if (ecb_expect_true (len <= 0xffU))
197	{	260	{
198	*enc->cur++ = major | 24;	261	*enc->cur++ = major | LENGTH_EXT1;
199	*enc->cur++ = len;	262	*enc->cur++ = len;
200	}	263	}
201	else if (len <= 0xffff)	264	else if (len <= 0xffffU)
202	{	265	{
203	*enc->cur++ = major | 25;	266	*enc->cur++ = major | LENGTH_EXT2;
204	*enc->cur++ = len >> 8;	267	*enc->cur++ = len >> 8;
205	*enc->cur++ = len;	268	*enc->cur++ = len;
206	}	269	}
207	else if (len <= 0xffffffff)	270	else if (len <= 0xffffffffU)
208	{	271	{
209	*enc->cur++ = major | 26;	272	*enc->cur++ = major | LENGTH_EXT4;
210	*enc->cur++ = len >> 24;	273	*enc->cur++ = len >> 24;
211	*enc->cur++ = len >> 16;	274	*enc->cur++ = len >> 16;
212	*enc->cur++ = len >> 8;	275	*enc->cur++ = len >> 8;
213	*enc->cur++ = len;	276	*enc->cur++ = len;
214	}	277	}
215	else	278	else
216	{	279	{
217	*enc->cur++ = major | 27;	280	*enc->cur++ = major | LENGTH_EXT8;
218	*enc->cur++ = len >> 56;	281	*enc->cur++ = len >> 56;
219	*enc->cur++ = len >> 48;	282	*enc->cur++ = len >> 48;
220	*enc->cur++ = len >> 40;	283	*enc->cur++ = len >> 40;
221	*enc->cur++ = len >> 32;	284	*enc->cur++ = len >> 32;
222	*enc->cur++ = len >> 24;	285	*enc->cur++ = len >> 24;
…		…
224	*enc->cur++ = len >> 8;	287	*enc->cur++ = len >> 8;
225	*enc->cur++ = len;	288	*enc->cur++ = len;
226	}	289	}
227	}	290	}
228		291
		292	// encodes a perl value into a CBOR integer
		293	ecb_inline void
		294	encode_int (enc_t *enc, SV *sv)
		295	{
		296	if (SvIsUV (sv))
		297	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		298	else if (SvIVX (sv) >= 0)
		299	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		300	else
		301	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		302	}
		303
229	ecb_inline void	304	ecb_inline void
230	encode_tag (enc_t *enc, UV tag)	305	encode_tag (enc_t *enc, UV tag)
231	{	306	{
232	encode_uint (enc, 0xc0, tag);	307	encode_uint (enc, MAJOR_TAG, tag);
233	}	308	}
234		309
235	static void	310	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		311	ecb_noinline static void
236	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	312	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237	{	313	{
		314	STRLEN ulen = len;
		315	U8 *p, *pend = (U8 *)str + len;
		316
		317	for (p = (U8 *)str; p < pend; ++p)
		318	ulen += *p >> 7; // count set high bits
		319
		320	encode_uint (enc, MAJOR_TEXT, ulen);
		321
		322	need (enc, ulen);
		323	for (p = (U8 *)str; p < pend; ++p)
		324	if (*p < 0x80)
		325	*enc->cur++ = *p;
		326	else
		327	{
		328	*enc->cur++ = 0xc0 + (*p >> 6);
		329	*enc->cur++ = 0x80 + (*p & 63);
		330	}
		331	}
		332
		333	ecb_inline void
		334	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		335	{
		336	if (ecb_expect_false (upgrade_utf8))
		337	if (!utf8)
		338	{
		339	encode_str_utf8 (enc, utf8, str, len);
		340	return;
		341	}
		342
		343	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		344	need (enc, len);
		345	memcpy (enc->cur, str, len);
		346	enc->cur += len;
		347	}
		348
		349	ecb_inline void
		350	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		351	{
238	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	352	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239	{	353	{
240	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	354	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241		355
242	if (SvOK (*svp))	356	if (SvOK (*svp))
243	{	357	{
244	// already registered, use stringref	358	// already registered, use stringref
245	encode_tag (enc, CBOR_TAG_STRINGREF);	359	encode_tag (enc, CBOR_TAG_STRINGREF);
246	encode_uint (enc, 0x00, SvUV (*svp));	360	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247	return;	361	return;
248	}	362	}
249	else if (len >= minimum_string_length (enc->stringref_idx))	363	else if (len >= minimum_string_length (enc->stringref_idx))
250	{	364	{
251	// register only	365	// register only
252	sv_setuv (*svp, enc->stringref_idx);	366	sv_setuv (*svp, enc->stringref_idx);
253	++enc->stringref_idx;	367	++enc->stringref_idx;
254	}	368	}
255	}	369	}
256		370
257	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	371	encode_str (enc, upgrade_utf8, utf8, str, len);
		372	}
		373
		374	ecb_inline void
		375	encode_float16 (enc_t *enc, NV nv)
		376	{
258	need (enc, len);	377	need (enc, 1+2);
		378
		379	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		380
		381	uint16_t fp = ecb_float_to_binary16 (nv);
		382
		383	if (!ecb_big_endian ())
		384	fp = ecb_bswap16 (fp);
		385
259	memcpy (enc->cur, str, len);	386	memcpy (enc->cur, &fp, 2);
260	enc->cur += len;	387	enc->cur += 2;
		388	}
		389
		390	ecb_inline void
		391	encode_float32 (enc_t *enc, NV nv)
		392	{
		393	need (enc, 1+4);
		394
		395	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		396
		397	uint32_t fp = ecb_float_to_binary32 (nv);
		398
		399	if (!ecb_big_endian ())
		400	fp = ecb_bswap32 (fp);
		401
		402	memcpy (enc->cur, &fp, 4);
		403	enc->cur += 4;
		404	}
		405
		406	ecb_inline void
		407	encode_float64 (enc_t *enc, NV nv)
		408	{
		409	need (enc, 1+8);
		410
		411	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		412
		413	uint64_t fp = ecb_double_to_binary64 (nv);
		414
		415	if (!ecb_big_endian ())
		416	fp = ecb_bswap64 (fp);
		417
		418	memcpy (enc->cur, &fp, 8);
		419	enc->cur += 8;
		420	}
		421
		422	ecb_inline void
		423	encode_forced (enc_t *enc, UV type, SV *sv)
		424	{
		425	switch (type)
		426	{
		427	case AS_CBOR:
		428	{
		429	STRLEN len;
		430	char *str = SvPVbyte (sv, len);
		431
		432	need (enc, len);
		433	memcpy (enc->cur, str, len);
		434	enc->cur += len;
		435	}
		436	break;
		437
		438	case AS_BYTES:
		439	{
		440	STRLEN len;
		441	char *str = SvPVbyte (sv, len);
		442	encode_strref (enc, 0, 0, str, len);
		443	}
		444	break;
		445
		446	case AS_TEXT:
		447	{
		448	STRLEN len;
		449	char *str = SvPVutf8 (sv, len);
		450	encode_strref (enc, 1, 1, str, len);
		451	}
		452	break;
		453
		454	case AS_INT: encode_int (enc, sv); break;
		455	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		456	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		457	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		458
		459	default:
		460	croak ("encountered malformed CBOR::XS::Tagged object");
		461	}
261	}	462	}
262		463
263	static void encode_sv (enc_t *enc, SV *sv);	464	static void encode_sv (enc_t *enc, SV *sv);
264		465
265	static void	466	static void
…		…
270	if (enc->depth >= enc->cbor.max_depth)	471	if (enc->depth >= enc->cbor.max_depth)
271	croak (ERR_NESTING_EXCEEDED);	472	croak (ERR_NESTING_EXCEEDED);
272		473
273	++enc->depth;	474	++enc->depth;
274		475
275	encode_uint (enc, 0x80, len + 1);	476	encode_uint (enc, MAJOR_ARRAY, len + 1);
276		477
		478	if (ecb_expect_false (SvMAGICAL (av)))
277	for (i = 0; i <= len; ++i)	479	for (i = 0; i <= len; ++i)
278	{	480	{
279	SV **svp = av_fetch (av, i, 0);	481	SV **svp = av_fetch (av, i, 0);
280	encode_sv (enc, svp ? *svp : &PL_sv_undef);	482	encode_sv (enc, svp ? *svp : &PL_sv_undef);
281	}	483	}
		484	else
		485	for (i = 0; i <= len; ++i)
		486	{
		487	SV *sv = AvARRAY (av)[i];
		488	encode_sv (enc, sv ? sv : &PL_sv_undef);
		489	}
282		490
283	--enc->depth;	491	--enc->depth;
284	}	492	}
285		493
286	static void	494	static void
…		…
294	++enc->depth;	502	++enc->depth;
295		503
296	int pairs = hv_iterinit (hv);	504	int pairs = hv_iterinit (hv);
297	int mg = SvMAGICAL (hv);	505	int mg = SvMAGICAL (hv);
298		506
299	if (mg)	507	if (ecb_expect_false (mg))
300	encode_ch (enc, 0xa0 | 31);	508	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301	else	509	else
302	encode_uint (enc, 0xa0, pairs);	510	encode_uint (enc, MAJOR_MAP, pairs);
303		511
304	while ((he = hv_iternext (hv)))	512	while ((he = hv_iternext (hv)))
305	{	513	{
306	if (HeKLEN (he) == HEf_SVKEY)	514	if (HeKLEN (he) == HEf_SVKEY)
307	encode_sv (enc, HeSVKEY (he));	515	encode_sv (enc, HeSVKEY (he));
308	else	516	else
309	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	517	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310		518
311	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	519	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312	}	520	}
313		521
314	if (mg)	522	if (ecb_expect_false (mg))
315	encode_ch (enc, 0xe0 | 31);	523	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316		524
317	--enc->depth;	525	--enc->depth;
318	}	526	}
319		527
320	// encode objects, arrays and special \0=false and \1=true values.	528	// encode objects, arrays and special \0=false and \1=true values.
321	static void	529	static void
322	encode_rv (enc_t *enc, SV *sv)	530	encode_rv (enc_t *enc, SV *sv)
323	{	531	{
324	SvGETMAGIC (sv);	532	SvGETMAGIC (sv);
325
326	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327	&& ecb_expect_false (SvREFCNT (sv) > 1))
328	{
329	if (!enc->shareable)
330	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334	if (SvOK (*svp))
335	{
336	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337	encode_uint (enc, 0x00, SvUV (*svp));
338	return;
339	}
340	else
341	{
342	sv_setuv (*svp, enc->shareable_idx);
343	++enc->shareable_idx;
344	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345	}
346	}
347		533
348	svtype svt = SvTYPE (sv);	534	svtype svt = SvTYPE (sv);
349		535
350	if (ecb_expect_false (SvOBJECT (sv)))	536	if (ecb_expect_false (SvOBJECT (sv)))
351	{	537	{
…		…
358	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	544	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359	? cbor_tagged_stash	545	? cbor_tagged_stash
360	: gv_stashpv ("CBOR::XS::Tagged" , 1);	546	: gv_stashpv ("CBOR::XS::Tagged" , 1);
361		547
362	HV *stash = SvSTASH (sv);	548	HV *stash = SvSTASH (sv);
363	GV *method;
364		549
365	if (stash == boolean_stash)	550	if (stash == boolean_stash)
366	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	551	{
		552	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		553	return;
		554	}
367	else if (stash == error_stash)	555	else if (stash == error_stash)
368	encode_ch (enc, 0xe0 | 23);	556	{
		557	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		558	return;
		559	}
369	else if (stash == tagged_stash)	560	else if (stash == tagged_stash)
370	{	561	{
371	if (svt != SVt_PVAV)	562	if (svt != SVt_PVAV)
372	croak ("encountered CBOR::XS::Tagged object that isn't an array");	563	croak ("encountered CBOR::XS::Tagged object that isn't an array");
373		564
		565	switch (av_len ((AV *)sv))
		566	{
		567	case 2-1:
		568	// actually a tagged value
374	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	569	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	570	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		571	break;
		572
		573	case 3-1:
		574	// a forced type [value, type, undef]
		575	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		576	break;
		577
		578	default:
		579	croak ("encountered malformed CBOR::XS::Tagged object");
		580	}
		581
		582	return;
		583	}
		584	}
		585
		586	if (ecb_expect_false (SvREFCNT (sv) > 1)
		587	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		588	{
		589	if (ecb_expect_false (!enc->shareable))
		590	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		591
		592	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		593
		594	if (SvOK (*svp))
376	}	595	{
		596	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		597	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		598	return;
		599	}
		600	else
		601	{
		602	sv_setuv (*svp, enc->shareable_idx);
		603	++enc->shareable_idx;
		604	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		605	}
		606	}
		607
		608	if (ecb_expect_false (SvOBJECT (sv)))
		609	{
		610	HV *stash = SvSTASH (sv);
		611	GV *method;
		612
		613	if (enc->cbor.flags & F_FORBID_OBJECTS)
		614	croak ("encountered object '%s', but forbid_objects is enabled",
		615	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
377	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	616	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378	{	617	{
379	dSP;	618	dSP;
380		619
381	ENTER; SAVETMPS; PUSHMARK (SP);	620	ENTER; SAVETMPS;
		621	PUSHMARK (SP);
382	// we re-bless the reference to get overload and other niceties right	622	// we re-bless the reference to get overload and other niceties right
383	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	623	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384		624
385	PUTBACK;	625	PUTBACK;
386	// G_SCALAR ensures that return value is 1	626	// G_SCALAR ensures that return value is 1
…		…
399	}	639	}
400	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	640	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401	{	641	{
402	dSP;	642	dSP;
403		643
404	ENTER; SAVETMPS; PUSHMARK (SP);	644	ENTER; SAVETMPS;
		645	PUSHMARK (SP);
405	EXTEND (SP, 2);	646	EXTEND (SP, 2);
406	// we re-bless the reference to get overload and other niceties right	647	// we re-bless the reference to get overload and other niceties right
407	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	648	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408	PUSHs (sv_cbor);	649	PUSHs (sv_cbor);
409		650
…		…
414	// catch this surprisingly common error	655	// catch this surprisingly common error
415	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	656	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	657	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417		658
418	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	659	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419	encode_uint (enc, 0x80, count + 1);	660	encode_uint (enc, MAJOR_ARRAY, count + 1);
420	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	661	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421		662
422	while (count)	663	{
		664	int i;
		665
		666	for (i = 0; i < count; ++i)
423	encode_sv (enc, SP[1 - count--]);	667	encode_sv (enc, SP[i + 1 - count]);
		668
		669	SP -= count;
		670	}
424		671
425	PUTBACK;	672	PUTBACK;
426		673
427	FREETMPS; LEAVE;	674	FREETMPS; LEAVE;
428	}	675	}
…		…
446	{	693	{
447	double nv = SvNVX (sv);	694	double nv = SvNVX (sv);
448		695
449	need (enc, 9);	696	need (enc, 9);
450		697
451	if (ecb_expect_false (nv == (U32)nv))	698	if (ecb_expect_false (nv == (NV)(U32)nv))
452	encode_uint (enc, 0x00, (U32)nv);	699	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453	//TODO: maybe I32?	700	//TODO: maybe I32?
454	else if (ecb_expect_false (nv == (float)nv))	701	else if (ecb_expect_false (nv == (float)nv))
455	{	702	encode_float32 (enc, nv);
456	uint32_t fp = ecb_float_to_binary32 (nv);
457
458	*enc->cur++ = 0xe0 | 26;
459
460	if (!ecb_big_endian ())
461	fp = ecb_bswap32 (fp);
462
463	memcpy (enc->cur, &fp, 4);
464	enc->cur += 4;
465	}
466	else	703	else
467	{	704	encode_float64 (enc, nv);
468	uint64_t fp = ecb_double_to_binary64 (nv);
469
470	*enc->cur++ = 0xe0 | 27;
471
472	if (!ecb_big_endian ())
473	fp = ecb_bswap64 (fp);
474
475	memcpy (enc->cur, &fp, 8);
476	enc->cur += 8;
477	}
478	}	705	}
479		706
480	static void	707	static void
481	encode_sv (enc_t *enc, SV *sv)	708	encode_sv (enc_t *enc, SV *sv)
482	{	709	{
…		…
484		711
485	if (SvPOKp (sv))	712	if (SvPOKp (sv))
486	{	713	{
487	STRLEN len;	714	STRLEN len;
488	char *str = SvPV (sv, len);	715	char *str = SvPV (sv, len);
489	encode_str (enc, SvUTF8 (sv), str, len);	716	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490	}	717	}
491	else if (SvNOKp (sv))	718	else if (SvNOKp (sv))
492	encode_nv (enc, sv);	719	encode_nv (enc, sv);
493	else if (SvIOKp (sv))	720	else if (SvIOKp (sv))
494	{	721	encode_int (enc, sv);
495	if (SvIsUV (sv))
496	encode_uint (enc, 0x00, SvUVX (sv));
497	else if (SvIVX (sv) >= 0)
498	encode_uint (enc, 0x00, SvIVX (sv));
499	else
500	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
501	}
502	else if (SvROK (sv))	722	else if (SvROK (sv))
503	encode_rv (enc, SvRV (sv));	723	encode_rv (enc, SvRV (sv));
504	else if (!SvOK (sv))	724	else if (!SvOK (sv))
505	encode_ch (enc, 0xe0 | 22);	725	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	726	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507	encode_ch (enc, 0xe0 | 23);	727	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508	else	728	else
509	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	729	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	730	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511	}	731	}
512		732
513	static SV *	733	static SV *
514	encode_cbor (SV *scalar, CBOR *cbor)	734	encode_cbor (SV *scalar, CBOR *cbor)
515	{	735	{
516	enc_t enc = { };	736	enc_t enc = { 0 };
517		737
518	enc.cbor = *cbor;	738	enc.cbor = *cbor;
519	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	739	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520	enc.cur = SvPVX (enc.sv);	740	enc.cur = SvPVX (enc.sv);
521	enc.end = SvEND (enc.sv);	741	enc.end = SvEND (enc.sv);
522		742
523	SvPOK_only (enc.sv);	743	SvPOK_only (enc.sv);
524		744
525	if (cbor->flags & F_ALLOW_STRINGREF)	745	if (cbor->flags & F_PACK_STRINGS)
526	{	746	{
527	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	747	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	748	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	749	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530	}	750	}
…		…
553	U32 depth; // recursion depth	773	U32 depth; // recursion depth
554	U32 maxdepth; // recursion depth limit	774	U32 maxdepth; // recursion depth limit
555	AV *shareable;	775	AV *shareable;
556	AV *stringref;	776	AV *stringref;
557	SV *decode_tagged;	777	SV *decode_tagged;
		778	SV *err_sv; // optional sv for error, needs to be freed
558	} dec_t;	779	} dec_t;
559		780
560	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	781	// set dec->err to ERRSV
		782	ecb_cold static void
		783	err_errsv (dec_t *dec)
		784	{
		785	if (!dec->err)
		786	{
		787	dec->err_sv = newSVsv (ERRSV);
561		788
562	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	789	// chop off the trailing \n
		790	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		791	*SvEND (dec->err_sv) = 0;
563		792
		793	dec->err = SvPVutf8_nolen (dec->err_sv);
		794	}
		795	}
		796
		797	// the following functions are used to reduce code size and help the compiler to optimise
		798	ecb_cold static void
		799	err_set (dec_t *dec, const char *reason)
		800	{
		801	if (!dec->err)
		802	dec->err = reason;
		803	}
		804
		805	ecb_cold static void
		806	err_unexpected_end (dec_t *dec)
		807	{
		808	err_set (dec, "unexpected end of CBOR data");
		809	}
		810
		811	#define ERR_DO(do) SB do; goto fail; SE
		812	#define ERR(reason) ERR_DO (err_set (dec, reason))
		813	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		814
		815	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		816
564	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	817	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
565	#define DEC_DEC_DEPTH --dec->depth	818	#define DEC_DEC_DEPTH --dec->depth
566		819
567	static UV	820	static UV
568	decode_uint (dec_t *dec)	821	decode_uint (dec_t *dec)
569	{	822	{
570	switch (*dec->cur & 31)	823	U8 m = *dec->cur & MINOR_MASK;
571	{	824	++dec->cur;
572	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575	return *dec->cur++ & 31;
576		825
577	case 24:	826	if (ecb_expect_true (m < LENGTH_EXT1))
		827	return m;
		828	else if (ecb_expect_true (m == LENGTH_EXT1))
		829	{
578	WANT (2);	830	WANT (1);
579	dec->cur += 2;	831	dec->cur += 1;
580	return dec->cur[-1];	832	return dec->cur[-1];
581		833	}
582	case 25:	834	else if (ecb_expect_true (m == LENGTH_EXT2))
		835	{
583	WANT (3);	836	WANT (2);
584	dec->cur += 3;	837	dec->cur += 2;
585	return (((UV)dec->cur[-2]) << 8)	838	return (((UV)dec->cur[-2]) << 8)
586	| ((UV)dec->cur[-1]);	839	| ((UV)dec->cur[-1]);
587		840	}
588	case 26:	841	else if (ecb_expect_true (m == LENGTH_EXT4))
		842	{
589	WANT (5);	843	WANT (4);
590	dec->cur += 5;	844	dec->cur += 4;
591	return (((UV)dec->cur[-4]) << 24)	845	return (((UV)dec->cur[-4]) << 24)
592	| (((UV)dec->cur[-3]) << 16)	846	| (((UV)dec->cur[-3]) << 16)
593	| (((UV)dec->cur[-2]) << 8)	847	| (((UV)dec->cur[-2]) << 8)
594	| ((UV)dec->cur[-1]);	848	| ((UV)dec->cur[-1]);
595		849	}
596	case 27:	850	else if (ecb_expect_true (m == LENGTH_EXT8))
		851	{
597	WANT (9);	852	WANT (8);
598	dec->cur += 9;	853	dec->cur += 8;
		854
		855	return
		856	#if UVSIZE < 8
		857	0
		858	#else
599	return (((UV)dec->cur[-8]) << 56)	859	(((UV)dec->cur[-8]) << 56)
600	| (((UV)dec->cur[-7]) << 48)	860	| (((UV)dec->cur[-7]) << 48)
601	| (((UV)dec->cur[-6]) << 40)	861	| (((UV)dec->cur[-6]) << 40)
602	| (((UV)dec->cur[-5]) << 32)	862	| (((UV)dec->cur[-5]) << 32)
		863	#endif
603	| (((UV)dec->cur[-4]) << 24)	864	| (((UV)dec->cur[-4]) << 24)
604	| (((UV)dec->cur[-3]) << 16)	865	| (((UV)dec->cur[-3]) << 16)
605	| (((UV)dec->cur[-2]) << 8)	866	| (((UV)dec->cur[-2]) << 8)
606	| ((UV)dec->cur[-1]);	867	| ((UV)dec->cur[-1]);
607		868	}
608	default:	869	else
609	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	870	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610	}
611		871
612	fail:	872	fail:
613	return 0;	873	return 0;
614	}	874	}
615		875
…		…
620	{	880	{
621	AV *av = newAV ();	881	AV *av = newAV ();
622		882
623	DEC_INC_DEPTH;	883	DEC_INC_DEPTH;
624		884
625	if ((*dec->cur & 31) == 31)	885	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626	{	886	{
627	++dec->cur;	887	++dec->cur;
628		888
629	for (;;)	889	for (;;)
630	{	890	{
631	WANT (1);	891	WANT (1);
632		892
633	if (*dec->cur == (0xe0 | 31))	893	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
634	{	894	{
635	++dec->cur;	895	++dec->cur;
636	break;	896	break;
637	}	897	}
638		898
639	av_push (av, decode_sv (dec));	899	av_push (av, decode_sv (dec));
640	}	900	}
641	}	901	}
642	else	902	else
643	{	903	{
644	int i, len = decode_uint (dec);	904	UV i, len = decode_uint (dec);
645		905
		906	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646	av_fill (av, len - 1);	907	av_fill (av, len - 1);
647		908
648	for (i = 0; i < len; ++i)	909	for (i = 0; i < len; ++i)
649	AvARRAY (av)[i] = decode_sv (dec);	910	AvARRAY (av)[i] = decode_sv (dec);
650	}	911	}
651		912
652	DEC_DEC_DEPTH;	913	DEC_DEC_DEPTH;
653	return newRV_noinc ((SV *)av);	914	return newRV_noinc ((SV *)av);
654		915
655	fail:	916	fail:
656	SvREFCNT_dec (av);	917	SvREFCNT_dec_NN (av);
657	DEC_DEC_DEPTH;	918	DEC_DEC_DEPTH;
658	return &PL_sv_undef;	919	return &PL_sv_undef;
659	}	920	}
660		921
661	static void	922	static void
…		…
663	{	924	{
664	// for speed reasons, we specialcase single-string	925	// for speed reasons, we specialcase single-string
665	// byte or utf-8 strings as keys, but only when !stringref	926	// byte or utf-8 strings as keys, but only when !stringref
666		927
667	if (ecb_expect_true (!dec->stringref))	928	if (ecb_expect_true (!dec->stringref))
668	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	929	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669	{	930	{
670	I32 len = decode_uint (dec);	931	STRLEN len = decode_uint (dec);
671	char *key = (char *)dec->cur;	932	char *key = (char *)dec->cur;
672		933
		934	WANT (len);
673	dec->cur += len;	935	dec->cur += len;
674		936
675	if (ecb_expect_false (dec->stringref))
676	av_push (dec->stringref, newSVpvn (key, len));
677
678	hv_store (hv, key, len, decode_sv (dec), 0);	937	hv_store (hv, key, len, decode_sv (dec), 0);
679		938
680	return;	939	return;
681	}	940	}
682	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	941	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683	{	942	{
684	I32 len = decode_uint (dec);	943	STRLEN len = decode_uint (dec);
685	char *key = (char *)dec->cur;	944	char *key = (char *)dec->cur;
686		945
		946	WANT (len);
687	dec->cur += len;	947	dec->cur += len;
688		948
689	if (ecb_expect_false (dec->stringref))	949	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	950	if (!is_utf8_string (key, len))
		951	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691		952
692	hv_store (hv, key, -len, decode_sv (dec), 0);	953	hv_store (hv, key, -len, decode_sv (dec), 0);
693		954
694	return;	955	return;
695	}	956	}
696		957
697	SV *k = decode_sv (dec);	958	SV *k = decode_sv (dec);
698	SV *v = decode_sv (dec);	959	SV *v = decode_sv (dec);
699		960
		961	// we leak memory if uncaught exceptions are thrown by random magical
		962	// methods, and this is hopefully the only place where it can happen,
		963	// so if there is a chance of an exception, take the very slow path.
		964	// since catching exceptions is "undocumented/internal/forbidden" by
		965	// the new p5p powers, we need to call out to a perl function :/
		966	if (ecb_expect_false (SvAMAGIC (k)))
		967	{
		968	dSP;
		969
		970	ENTER; SAVETMPS;
		971	PUSHMARK (SP);
		972	EXTEND (SP, 3);
		973	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		974	PUSHs (sv_2mortal (k));
		975	PUSHs (sv_2mortal (v));
		976
		977	PUTBACK;
		978	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		979	SPAGAIN;
		980
		981	FREETMPS; LEAVE;
		982
		983	if (SvTRUE (ERRSV))
		984	ERR_ERRSV;
		985
		986	return;
		987	}
		988
700	hv_store_ent (hv, k, v, 0);	989	hv_store_ent (hv, k, v, 0);
701	SvREFCNT_dec (k);	990	SvREFCNT_dec_NN (k);
		991
		992	fail:
		993	;
702	}	994	}
703		995
704	static SV *	996	static SV *
705	decode_hv (dec_t *dec)	997	decode_hv (dec_t *dec)
706	{	998	{
707	HV *hv = newHV ();	999	HV *hv = newHV ();
708		1000
709	DEC_INC_DEPTH;	1001	DEC_INC_DEPTH;
710		1002
711	if ((*dec->cur & 31) == 31)	1003	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712	{	1004	{
713	++dec->cur;	1005	++dec->cur;
714		1006
715	for (;;)	1007	for (;;)
716	{	1008	{
717	WANT (1);	1009	WANT (1);
718		1010
719	if (*dec->cur == (0xe0 | 31))	1011	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
720	{	1012	{
721	++dec->cur;	1013	++dec->cur;
722	break;	1014	break;
723	}	1015	}
724		1016
725	decode_he (dec, hv);	1017	decode_he (dec, hv);
726	}	1018	}
727	}	1019	}
728	else	1020	else
729	{	1021	{
730	int pairs = decode_uint (dec);	1022	UV pairs = decode_uint (dec);
		1023
		1024	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731		1025
732	while (pairs--)	1026	while (pairs--)
733	decode_he (dec, hv);	1027	decode_he (dec, hv);
734	}	1028	}
735		1029
736	DEC_DEC_DEPTH;	1030	DEC_DEC_DEPTH;
737	return newRV_noinc ((SV *)hv);	1031	return newRV_noinc ((SV *)hv);
738		1032
739	fail:	1033	fail:
740	SvREFCNT_dec (hv);	1034	SvREFCNT_dec_NN (hv);
741	DEC_DEC_DEPTH;	1035	DEC_DEC_DEPTH;
742	return &PL_sv_undef;	1036	return &PL_sv_undef;
743	}	1037	}
744		1038
745	static SV *	1039	static SV *
746	decode_str (dec_t *dec, int utf8)	1040	decode_str (dec_t *dec, int utf8)
747	{	1041	{
748	SV *sv = 0;	1042	SV *sv = 0;
749		1043
750	if ((*dec->cur & 31) == 31)	1044	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
751	{	1045	{
		1046	// indefinite length strings
752	++dec->cur;	1047	++dec->cur;
753		1048
		1049	U8 major = *dec->cur & MAJOR_MISC;
		1050
754	sv = newSVpvn ("", 0);	1051	sv = newSVpvn ("", 0);
755		1052
756	// not very fast, and certainly not robust against illegal input
757	for (;;)	1053	for (;;)
758	{	1054	{
759	WANT (1);	1055	WANT (1);
760		1056
761	if (*dec->cur == (0xe0 | 31))	1057	if ((*dec->cur - major) > LENGTH_EXT8)
		1058	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762	{	1059	{
763	++dec->cur;	1060	++dec->cur;
764	break;	1061	break;
765	}	1062	}
		1063	else
		1064	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766		1065
767	sv_catsv (sv, decode_sv (dec));	1066	STRLEN len = decode_uint (dec);
		1067
		1068	WANT (len);
		1069	sv_catpvn (sv, dec->cur, len);
		1070	dec->cur += len;
768	}	1071	}
769	}	1072	}
770	else	1073	else
771	{	1074	{
772	STRLEN len = decode_uint (dec);	1075	STRLEN len = decode_uint (dec);
…		…
779	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1082	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1083	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781	}	1084	}
782		1085
783	if (utf8)	1086	if (utf8)
		1087	{
		1088	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1089	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1090	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1091
784	SvUTF8_on (sv);	1092	SvUTF8_on (sv);
		1093	}
785		1094
786	return sv;	1095	return sv;
787		1096
788	fail:	1097	fail:
789	SvREFCNT_dec (sv);	1098	SvREFCNT_dec (sv);
…		…
808	sv = newRV_noinc (decode_sv (dec));	1117	sv = newRV_noinc (decode_sv (dec));
809	break;	1118	break;
810		1119
811	case CBOR_TAG_STRINGREF_NAMESPACE:	1120	case CBOR_TAG_STRINGREF_NAMESPACE:
812	{	1121	{
		1122	// do not use SAVETMPS/FREETMPS, as these will
		1123	// erase mortalised caches, e.g. "shareable"
813	ENTER; SAVETMPS;	1124	ENTER;
814		1125
815	SAVESPTR (dec->stringref);	1126	SAVESPTR (dec->stringref);
816	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1127	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817		1128
818	sv = decode_sv (dec);	1129	sv = decode_sv (dec);
819		1130
820	FREETMPS; LEAVE;	1131	LEAVE;
821	}	1132	}
822	break;	1133	break;
823		1134
824	case CBOR_TAG_STRINGREF:	1135	case CBOR_TAG_STRINGREF:
825	{	1136	{
826	if ((*dec->cur >> 5) != 0)	1137	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1138	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828		1139
829	UV idx = decode_uint (dec);	1140	UV idx = decode_uint (dec);
830		1141
831	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1142	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
832	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1143	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
833		1144
834	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1145	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
835	}	1146	}
836	break;	1147	break;
…		…
838	case CBOR_TAG_VALUE_SHAREABLE:	1149	case CBOR_TAG_VALUE_SHAREABLE:
839	{	1150	{
840	if (ecb_expect_false (!dec->shareable))	1151	if (ecb_expect_false (!dec->shareable))
841	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1152	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842		1153
		1154	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1155	{
843	sv = newSV (0);	1156	sv = newSV (0);
844	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1157	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845		1158
846	SV *osv = decode_sv (dec);	1159	SV *osv = decode_sv (dec);
847	sv_setsv (sv, osv);	1160	sv_setsv (sv, osv);
848	SvREFCNT_dec_NN (osv);	1161	SvREFCNT_dec_NN (osv);
		1162	}
		1163	else
		1164	{
		1165	av_push (dec->shareable, &PL_sv_undef);
		1166	int idx = AvFILLp (dec->shareable);
		1167	sv = decode_sv (dec);
		1168	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1169	}
849	}	1170	}
850	break;	1171	break;
851		1172
852	case CBOR_TAG_VALUE_SHAREDREF:	1173	case CBOR_TAG_VALUE_SHAREDREF:
853	{	1174	{
854	if ((*dec->cur >> 5) != 0)	1175	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1176	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856		1177
857	UV idx = decode_uint (dec);	1178	UV idx = decode_uint (dec);
858		1179
859	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1180	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
860	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1181	ERR ("corrupted CBOR data (sharedref index out of bounds)");
861		1182
862	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1183	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1184
		1185	if (sv == &PL_sv_undef)
		1186	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863	}	1187	}
864	break;	1188	break;
865		1189
866	case CBOR_TAG_PERL_OBJECT:	1190	case CBOR_TAG_PERL_OBJECT:
867	{	1191	{
		1192	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1193	goto filter;
		1194
868	sv = decode_sv (dec);	1195	sv = decode_sv (dec);
869		1196
870	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1197	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871	ERR ("corrupted CBOR data (non-array perl object)");	1198	ERR ("corrupted CBOR data (non-array perl object)");
872		1199
…		…
882	if (!method)	1209	if (!method)
883	ERR ("cannot decode perl-object (package does not have a THAW method)");	1210	ERR ("cannot decode perl-object (package does not have a THAW method)");
884		1211
885	dSP;	1212	dSP;
886		1213
887	ENTER; SAVETMPS; PUSHMARK (SP);	1214	ENTER; SAVETMPS;
		1215	PUSHMARK (SP);
888	EXTEND (SP, len + 1);	1216	EXTEND (SP, len + 1);
889	// we re-bless the reference to get overload and other niceties right	1217	// we re-bless the reference to get overload and other niceties right
890	PUSHs (*av_fetch (av, 0, 1));	1218	PUSHs (*av_fetch (av, 0, 1));
891	PUSHs (sv_cbor);	1219	PUSHs (sv_cbor);
892		1220
…		…
900	SPAGAIN;	1228	SPAGAIN;
901		1229
902	if (SvTRUE (ERRSV))	1230	if (SvTRUE (ERRSV))
903	{	1231	{
904	FREETMPS; LEAVE;	1232	FREETMPS; LEAVE;
905	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1233	ERR_ERRSV;
906	}	1234	}
907		1235
908	SvREFCNT_dec (sv);	1236	SvREFCNT_dec_NN (sv);
909	sv = SvREFCNT_inc (POPs);	1237	sv = SvREFCNT_inc (POPs);
910		1238
911	PUTBACK;	1239	PUTBACK;
912		1240
913	FREETMPS; LEAVE;	1241	FREETMPS; LEAVE;
914	}	1242	}
915	break;	1243	break;
916		1244
917	default:	1245	default:
		1246	filter:
918	{	1247	{
		1248	SV *tag_sv = newSVuv (tag);
		1249
919	sv = decode_sv (dec);	1250	sv = decode_sv (dec);
920		1251
921	dSP;	1252	dSP;
922	ENTER; SAVETMPS; PUSHMARK (SP);	1253	ENTER; SAVETMPS;
		1254	PUSHMARK (SP);
923	EXTEND (SP, 2);	1255	EXTEND (SP, 2);
924	PUSHs (newSVuv (tag));	1256	PUSHs (tag_sv);
925	PUSHs (sv);	1257	PUSHs (sv);
926		1258
927	PUTBACK;	1259	PUTBACK;
928	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1260	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929	SPAGAIN;	1261	SPAGAIN;
930		1262
931	if (SvTRUE (ERRSV))	1263	if (SvTRUE (ERRSV))
932	{	1264	{
		1265	SvREFCNT_dec_NN (tag_sv);
933	FREETMPS; LEAVE;	1266	FREETMPS; LEAVE;
934	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1267	ERR_ERRSV;
935	}	1268	}
936		1269
937	if (count)	1270	if (count)
938	{	1271	{
		1272	SvREFCNT_dec_NN (tag_sv);
939	SvREFCNT_dec (sv);	1273	SvREFCNT_dec_NN (sv);
940	sv = SvREFCNT_inc (POPs);	1274	sv = SvREFCNT_inc_NN (TOPs);
		1275	SP -= count;
941	}	1276	}
942	else	1277	else
943	{	1278	{
944	AV *av = newAV ();	1279	AV *av = newAV ();
945	av_push (av, newSVuv (tag));	1280	av_push (av, tag_sv);
946	av_push (av, sv);	1281	av_push (av, sv);
947		1282
948	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1283	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949	? cbor_tagged_stash	1284	? cbor_tagged_stash
950	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1285	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
968	static SV *	1303	static SV *
969	decode_sv (dec_t *dec)	1304	decode_sv (dec_t *dec)
970	{	1305	{
971	WANT (1);	1306	WANT (1);
972		1307
973	switch (*dec->cur >> 5)	1308	switch (*dec->cur >> MAJOR_SHIFT)
974	{	1309	{
975	case 0: // unsigned int	1310	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976	return newSVuv (decode_uint (dec));	1311	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977	case 1: // negative int	1312	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978	return newSViv (-1 - (IV)decode_uint (dec));	1313	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979	case 2: // octet string	1314	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980	return decode_str (dec, 0);	1315	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981	case 3: // utf-8 string	1316	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982	return decode_str (dec, 1);	1317
983	case 4: // array	1318	case MAJOR_MISC >> MAJOR_SHIFT:
984	return decode_av (dec);
985	case 5: // map
986	return decode_hv (dec);
987	case 6: // tag
988	return decode_tagged (dec);
989	case 7: // misc
990	switch (*dec->cur++ & 31)	1319	switch (*dec->cur++ & MINOR_MASK)
991	{	1320	{
992	case 20:	1321	case SIMPLE_FALSE:
993	#if CBOR_SLOW	1322	#if CBOR_SLOW
994	types_false = get_bool ("Types::Serialiser::false");	1323	types_false = get_bool ("Types::Serialiser::false");
995	#endif	1324	#endif
996	return newSVsv (types_false);	1325	return newSVsv (types_false);
997	case 21:	1326	case SIMPLE_TRUE:
998	#if CBOR_SLOW	1327	#if CBOR_SLOW
999	types_true = get_bool ("Types::Serialiser::true");	1328	types_true = get_bool ("Types::Serialiser::true");
1000	#endif	1329	#endif
1001	return newSVsv (types_true);	1330	return newSVsv (types_true);
1002	case 22:	1331	case SIMPLE_NULL:
1003	return newSVsv (&PL_sv_undef);	1332	return newSVsv (&PL_sv_undef);
1004	case 23:	1333	case SIMPLE_UNDEF:
1005	#if CBOR_SLOW	1334	#if CBOR_SLOW
1006	types_error = get_bool ("Types::Serialiser::error");	1335	types_error = get_bool ("Types::Serialiser::error");
1007	#endif	1336	#endif
1008	return newSVsv (types_error);	1337	return newSVsv (types_error);
1009		1338
1010	case 25:	1339	case MISC_FLOAT16:
1011	{	1340	{
1012	WANT (2);	1341	WANT (2);
1013		1342
1014	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1343	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015	dec->cur += 2;	1344	dec->cur += 2;
1016		1345
1017	return newSVnv (ecb_binary16_to_float (fp));	1346	return newSVnv (ecb_binary16_to_float (fp));
1018	}	1347	}
1019		1348
1020	case 26:	1349	case MISC_FLOAT32:
1021	{	1350	{
1022	uint32_t fp;	1351	uint32_t fp;
1023	WANT (4);	1352	WANT (4);
1024	memcpy (&fp, dec->cur, 4);	1353	memcpy (&fp, dec->cur, 4);
1025	dec->cur += 4;	1354	dec->cur += 4;
…		…
1028	fp = ecb_bswap32 (fp);	1357	fp = ecb_bswap32 (fp);
1029		1358
1030	return newSVnv (ecb_binary32_to_float (fp));	1359	return newSVnv (ecb_binary32_to_float (fp));
1031	}	1360	}
1032		1361
1033	case 27:	1362	case MISC_FLOAT64:
1034	{	1363	{
1035	uint64_t fp;	1364	uint64_t fp;
1036	WANT (8);	1365	WANT (8);
1037	memcpy (&fp, dec->cur, 8);	1366	memcpy (&fp, dec->cur, 8);
1038	dec->cur += 8;	1367	dec->cur += 8;
…		…
1041	fp = ecb_bswap64 (fp);	1370	fp = ecb_bswap64 (fp);
1042		1371
1043	return newSVnv (ecb_binary64_to_double (fp));	1372	return newSVnv (ecb_binary64_to_double (fp));
1044	}	1373	}
1045		1374
1046	// 0..19 unassigned	1375	// 0..19 unassigned simple
1047	// 24 reserved + unassigned (reserved values are not encodable)	1376	// 24 reserved + unassigned simple (reserved values are not encodable)
		1377	// 28-30 unassigned misc
		1378	// 31 break code
1048	default:	1379	default:
1049	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1380	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050	}	1381	}
1051		1382
1052	break;	1383	break;
1053	}	1384	}
1054		1385
…		…
1057	}	1388	}
1058		1389
1059	static SV *	1390	static SV *
1060	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1391	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061	{	1392	{
1062	dec_t dec = { };	1393	dec_t dec = { 0 };
1063	SV *sv;	1394	SV *sv;
1064	STRLEN len;	1395	STRLEN len;
1065	char *data = SvPVbyte (string, len);	1396	char *data = SvPVbyte (string, len);
1066		1397
1067	if (len > cbor->max_size && cbor->max_size)	1398	if (len > cbor->max_size && cbor->max_size)
…		…
1081	if (dec.cur != dec.end && !dec.err)	1412	if (dec.cur != dec.end && !dec.err)
1082	dec.err = "garbage after CBOR object";	1413	dec.err = "garbage after CBOR object";
1083		1414
1084	if (dec.err)	1415	if (dec.err)
1085	{	1416	{
		1417	if (dec.shareable)
		1418	{
		1419	// need to break cyclic links, which would all be in shareable
		1420	int i;
		1421	SV **svp;
		1422
		1423	for (i = av_len (dec.shareable) + 1; i--; )
		1424	if ((svp = av_fetch (dec.shareable, i, 0)))
		1425	sv_setsv (*svp, &PL_sv_undef);
		1426	}
		1427
1086	SvREFCNT_dec (sv);	1428	SvREFCNT_dec_NN (sv);
		1429
		1430	if (dec.err_sv)
		1431	sv_2mortal (dec.err_sv);
		1432
1087	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1433	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088	}	1434	}
1089		1435
1090	sv = sv_2mortal (sv);	1436	sv = sv_2mortal (sv);
1091		1437
1092	return sv;	1438	return sv;
1093	}	1439	}
1094		1440
		1441	/////////////////////////////////////////////////////////////////////////////
		1442	// incremental parser
		1443
		1444	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1445
		1446	// returns 0 for notyet, 1 for success or error
		1447	static int
		1448	incr_parse (CBOR *self, SV *cborstr)
		1449	{
		1450	STRLEN cur;
		1451	SvPV (cborstr, cur);
		1452
		1453	while (ecb_expect_true (self->incr_need <= cur))
		1454	{
		1455	// table of integer count bytes
		1456	static I8 incr_len[MINOR_MASK + 1] = {
		1457	0, 0, 0, 0, 0, 0, 0, 0,
		1458	0, 0, 0, 0, 0, 0, 0, 0,
		1459	0, 0, 0, 0, 0, 0, 0, 0,
		1460	1, 2, 4, 8,-1,-1,-1,-2
		1461	};
		1462
		1463	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1464	U8 m = *p & MINOR_MASK;
		1465	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1466	I8 ilen = incr_len[m];
		1467
		1468	self->incr_need = self->incr_pos + 1;
		1469
		1470	if (ecb_expect_false (ilen < 0))
		1471	{
		1472	if (m != MINOR_INDEF)
		1473	return 1; // error
		1474
		1475	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1476	{
		1477	if (count >= 0)
		1478	return 1; // error
		1479
		1480	count = 1;
		1481	}
		1482	else
		1483	{
		1484	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1485	count = -1;
		1486	}
		1487	}
		1488	else
		1489	{
		1490	self->incr_need += ilen;
		1491	if (ecb_expect_false (self->incr_need > cur))
		1492	return 0;
		1493
		1494	int major = *p >> MAJOR_SHIFT;
		1495
		1496	switch (major)
		1497	{
		1498	case MAJOR_TAG >> MAJOR_SHIFT:
		1499	++count; // tags merely prefix another value
		1500	break;
		1501
		1502	case MAJOR_BYTES >> MAJOR_SHIFT:
		1503	case MAJOR_TEXT >> MAJOR_SHIFT:
		1504	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1505	case MAJOR_MAP >> MAJOR_SHIFT:
		1506	{
		1507	UV len;
		1508
		1509	if (ecb_expect_false (ilen))
		1510	{
		1511	len = 0;
		1512
		1513	do {
		1514	len = (len << 8) | *++p;
		1515	} while (--ilen);
		1516	}
		1517	else
		1518	len = m;
		1519
		1520	switch (major)
		1521	{
		1522	case MAJOR_BYTES >> MAJOR_SHIFT:
		1523	case MAJOR_TEXT >> MAJOR_SHIFT:
		1524	self->incr_need += len;
		1525	if (ecb_expect_false (self->incr_need > cur))
		1526	return 0;
		1527
		1528	break;
		1529
		1530	case MAJOR_MAP >> MAJOR_SHIFT:
		1531	len <<= 1;
		1532	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1533	if (len)
		1534	{
		1535	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1536	count = len + 1;
		1537	}
		1538	break;
		1539	}
		1540	}
		1541	}
		1542	}
		1543
		1544	self->incr_pos = self->incr_need;
		1545
		1546	if (count > 0)
		1547	{
		1548	while (!--count)
		1549	{
		1550	if (!AvFILLp (self->incr_count))
		1551	return 1; // done
		1552
		1553	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1554	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1555	}
		1556
		1557	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1558	}
		1559	}
		1560
		1561	return 0;
		1562	}
		1563
		1564
1095	/////////////////////////////////////////////////////////////////////////////	1565	/////////////////////////////////////////////////////////////////////////////
1096	// XS interface functions	1566	// XS interface functions
1097		1567
1098	MODULE = CBOR::XS PACKAGE = CBOR::XS	1568	MODULE = CBOR::XS PACKAGE = CBOR::XS
1099		1569
…		…
1111		1581
1112	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1582	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113		1583
1114	sv_cbor = newSVpv ("CBOR", 0);	1584	sv_cbor = newSVpv ("CBOR", 0);
1115	SvREADONLY_on (sv_cbor);	1585	SvREADONLY_on (sv_cbor);
		1586
		1587	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116	}	1588	}
1117		1589
1118	PROTOTYPES: DISABLE	1590	PROTOTYPES: DISABLE
1119		1591
1120	void CLONE (...)	1592	void CLONE (...)
…		…
1139	void shrink (CBOR *self, int enable = 1)	1611	void shrink (CBOR *self, int enable = 1)
1140	ALIAS:	1612	ALIAS:
1141	shrink = F_SHRINK	1613	shrink = F_SHRINK
1142	allow_unknown = F_ALLOW_UNKNOWN	1614	allow_unknown = F_ALLOW_UNKNOWN
1143	allow_sharing = F_ALLOW_SHARING	1615	allow_sharing = F_ALLOW_SHARING
1144	allow_stringref = F_ALLOW_STRINGREF	1616	allow_cycles = F_ALLOW_CYCLES
		1617	forbid_objects = F_FORBID_OBJECTS
		1618	pack_strings = F_PACK_STRINGS
		1619	text_keys = F_TEXT_KEYS
		1620	text_strings = F_TEXT_STRINGS
		1621	validate_utf8 = F_VALIDATE_UTF8
1145	PPCODE:	1622	PPCODE:
1146	{	1623	{
1147	if (enable)	1624	if (enable)
1148	self->flags |= ix;	1625	self->flags |= ix;
1149	else	1626	else
…		…
1155	void get_shrink (CBOR *self)	1632	void get_shrink (CBOR *self)
1156	ALIAS:	1633	ALIAS:
1157	get_shrink = F_SHRINK	1634	get_shrink = F_SHRINK
1158	get_allow_unknown = F_ALLOW_UNKNOWN	1635	get_allow_unknown = F_ALLOW_UNKNOWN
1159	get_allow_sharing = F_ALLOW_SHARING	1636	get_allow_sharing = F_ALLOW_SHARING
1160	get_allow_stringref = F_ALLOW_STRINGREF	1637	get_allow_cycles = F_ALLOW_CYCLES
		1638	get_forbid_objects = F_FORBID_OBJECTS
		1639	get_pack_strings = F_PACK_STRINGS
		1640	get_text_keys = F_TEXT_KEYS
		1641	get_text_strings = F_TEXT_STRINGS
		1642	get_validate_utf8 = F_VALIDATE_UTF8
1161	PPCODE:	1643	PPCODE:
1162	XPUSHs (boolSV (self->flags & ix));	1644	XPUSHs (boolSV (self->flags & ix));
1163		1645
1164	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1646	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165	PPCODE:	1647	PPCODE:
…		…
1214	EXTEND (SP, 2);	1696	EXTEND (SP, 2);
1215	PUSHs (sv);	1697	PUSHs (sv);
1216	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1698	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217	}	1699	}
1218		1700
		1701	void incr_parse (CBOR *self, SV *cborstr)
		1702	ALIAS:
		1703	incr_parse_multiple = 1
		1704	PPCODE:
		1705	{
		1706	if (SvUTF8 (cborstr))
		1707	sv_utf8_downgrade (cborstr, 0);
		1708
		1709	if (!self->incr_count)
		1710	{
		1711	self->incr_count = newAV ();
		1712	self->incr_pos = 0;
		1713	self->incr_need = 1;
		1714
		1715	av_push (self->incr_count, newSViv (1));
		1716	}
		1717
		1718	do
		1719	{
		1720	if (!incr_parse (self, cborstr))
		1721	{
		1722	if (self->incr_need > self->max_size && self->max_size)
		1723	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1724	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1725
		1726	break;
		1727	}
		1728
		1729	SV *sv;
		1730	char *offset;
		1731
		1732	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1733	XPUSHs (sv);
		1734
		1735	sv_chop (cborstr, offset);
		1736
		1737	av_clear (self->incr_count);
		1738	av_push (self->incr_count, newSViv (1));
		1739
		1740	self->incr_pos = 0;
		1741	self->incr_need = self->incr_pos + 1;
		1742	}
		1743	while (ix);
		1744	}
		1745
		1746	void incr_reset (CBOR *self)
		1747	CODE:
		1748	{
		1749	SvREFCNT_dec (self->incr_count);
		1750	self->incr_count = 0;
		1751	}
		1752
1219	void DESTROY (CBOR *self)	1753	void DESTROY (CBOR *self)
1220	PPCODE:	1754	PPCODE:
1221	cbor_free (self);	1755	cbor_free (self);
1222		1756
1223	PROTOTYPES: ENABLE	1757	PROTOTYPES: ENABLE
1224		1758
1225	void encode_cbor (SV *scalar)	1759	void encode_cbor (SV *scalar)
		1760	ALIAS:
		1761	encode_cbor = 0
		1762	encode_cbor_sharing = F_ALLOW_SHARING
1226	PPCODE:	1763	PPCODE:
1227	{	1764	{
1228	CBOR cbor;	1765	CBOR cbor;
1229	cbor_init (&cbor);	1766	cbor_init (&cbor);
		1767	cbor.flags |= ix;
1230	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1768	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231	XPUSHs (scalar);	1769	XPUSHs (scalar);
1232	}	1770	}
1233		1771
1234	void decode_cbor (SV *cborstr)	1772	void decode_cbor (SV *cborstr)
…		…
1238	cbor_init (&cbor);	1776	cbor_init (&cbor);
1239	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1777	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1240	XPUSHs (cborstr);	1778	XPUSHs (cborstr);
1241	}	1779	}
1242		1780
		1781	#ifdef __AFL_COMPILER
		1782
		1783	void
		1784	afl_init ()
		1785	CODE:
		1786	__AFL_INIT ();
		1787
		1788	int
		1789	afl_loop (unsigned int count = 10000)
		1790	CODE:
		1791	RETVAL = __AFL_LOOP (count);
		1792	OUTPUT:
		1793	RETVAL
		1794
		1795	#endif
		1796

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines